Treasury meets AI: Insights for smarter decisions

We were recently invited by the Treasury Management Association of New York (TMANY) and the North Carolina Treasury Management Association (NCTMA) to deliver the online session for their members. In this webinar, we provided them with a clear framework to assess whether AI aligns with their treasury needs and to avoid common implementation pitfalls — from data readiness to stakeholder alignment.

The session was led by:

• Tanya Kohen, CTP — Treasury transformation expert with 20+ years of experience.

• Paul Chayka — Integration and AI Solution expert at WaveAccess.

• Neil Smullian — Head of Growth at WaveAccess.

Below, we've distilled the key highlights and actionable lessons from the webinar.

What is AI in Treasury?

At its core, AI is about teaching machines to perform tasks that typically require human intelligence. Think of it as a way to automate decision-making, recognize patterns, and predict outcomes — all based on data.

But here’s the thing: AI isn’t a single technology. It’s a collection of tools and techniques. Machine Learning, Deep Learning, Natural Language Processing and way more are all part of the AI family. These tools can be applied in different ways — whether it’s automating repetitive tasks, analyzing complex datasets, or understanding human language.

AI isn’t magic. It’s not about replacing humans; it’s about enhancing what we can do. In treasury, AI can process vast amounts of financial data quickly and accurately, helping you make smarter, faster decisions. But — and this is important — it’s only as good as the data and expertise you bring to the table.

Modern corporate finance is increasingly leveraging AI to boost efficiency and uncover valuable insights. Let’s look at three areas where AI is making a big impact:

1. Cash flow forecasting

   Machine Learning algorithms excel at analyzing historical data to predict future cash flows. By identifying patterns and trends, AI can provide more accurate forecasts, helping you make better-informed decisions about liquidity and investments.

2. AI-driven anomaly detection

   AI is incredibly effective at spotting anomalies in financial transactions. By learning what “normal” looks like, it can flag unusual activity in real time, reducing the risk of fraud and saving your organization significant time and money.

3. NLP-based AI assistants

   AI assistants, often using Natural Language Processing, can review contracts and payment terms quickly and accurately. This not only speeds up the process but also reduces the risk of human error, ensuring compliance and efficiency.

These are just a few examples, but they show how AI can be a powerful ally in treasury — when used the right way.

What do you need to know to implement AI

AI is not a one-size-fits-all solution, and it’s not easy to implement. Here’s what you need to know to approach AI effectively:

1. AI isn’t plug-and-play.

   In most cases, it’s not something you can just install and expect to work perfectly right away. Custom AI needs training, fine-tuning, and — importantly — human oversight to succeed. And let’s be clear: it won’t replace strategic human judgment. Think of it as a tool, not a replacement.

2. AI excels in specific scenarios, but its success hinges on foundational requirements:
   • Where AI delivers value is repetitive, data-heavy tasks. For example, AI-powered reconciliation tools can match thousands of transactions across bank accounts in seconds, reducing errors and manual work.
   • AI works best with clean, structured data and high-volume processes.

   For AI to work in these scenarios, three things are non-negotiable:

   • Quality data: AI’s accuracy depends entirely on the data it’s trained on. Garbage in, garbage out.
   • Seamless workflow integration: AI must align with your existing processes.
   • Clear objectives: Understanding the “why” behind AI adoption is critical. Define the business pain point to address upfront, like “We need to reduce reconciliation errors by 30%”. Without clear goals, AI becomes an expensive experiment.

   It’s also worth noting that not everything marketed as AI truly is. Traditional rule-based automation, business intelligence tools, and basic scripting are often mistaken for AI. While these tools are useful, they lack the adaptive learning capabilities that define true AI solutions.

3. There’s no point in implementing AI just for the sake of it. When does skipping AI make sense? 

   If your process has clear, predictable steps, simpler non-AI tools may be better. For instance, automating vendor payments using predefined rules is often faster and cheaper with Robotic Process Automation than with AI that requires continuous retraining.

   In our experience, other non-AI tools, technologies and approaches can successfully cope with a number of automation tasks in treasury.

   If you’re unsure, seek expert advice. Consulting with tech-savvy teams can help you choose the right solution for your needs.

NCTMA webinar recording

Common mistakes and how to avoid them

Now that we’ve covered where and when AI can add value, let’s talk about the most common mistakes and misconceptions on the path to successful AI implementation, and discuss how to avoid them. Fortunately, many of the pitfalls are predictable and, therefore, avoidable.

The bottom line is: AI is a powerful tool, but it’s not a one-size-fits-all solution. By understanding the strengths of both AI and traditional methods, you can choose the right approach for your specific needs.

In the next part, we’ll explore real-world examples of where AI is making a measurable impact in treasury — and the key requirements for success. You’ll see under what conditions AI works best and when standard automation methods might be the better choice for your processes.